The present invention relates generally to magnetic recording media used in rigid disc drives commonly used for computer data storage and methods of making the same.
In the hard disk drive industry, the ever-increasing recording density demands continuous improvement in hard disk recording media so as to support a higher linear recording density (thousands of flux changes per inchxe2x80x94KFCI) and track density (thousands of tracks per inchxe2x80x94KTPI). Recording density is proportional to the product of KFCI and KTPI, and is typically expressed as giga-bits per square inch (Gb/in2). Currently the recording density is increasing at compound annual growth rate of 60%.
In order for the media to be able to support the high KFCI (e.g., over 200 KFCI), the pulse width (PW50xe2x80x94pulse width at 50% of pulse amplitude) needs to be as small as possible to reduce the inter-symbol interference so that high resolution at high recording density can be obtained. The resolution is defined as the pulse amplitude at high frequency divided by the pulse amplitude at low frequency. Based on generally known magnetic recording theory, in order to reduce PW50 and hence increase resolution, the magnetic recording media must have high coercivity, Hc. For today""s typical recording density of 2 Gb/in2, the value of Hc needs to be on the order of 2500 Oe, and in future it needs to be 3000 Oe or more for even higher recording density. Other means of reducing PW50 include increasing the hysteresis loop squareness, generally defined as xe2x80x9cSxe2x80x9d which is ratio of remanent to saturation magnetization (Mr/Ms), increasing the coercivity squareness xe2x80x9cS*xe2x80x9d, increasing the remanent coercivity squareness xe2x80x9cS*remxe2x80x9d, and narrowing the switching field distribution (xe2x80x9cSFDxe2x80x9d), as described by William and Comstock in xe2x80x9cAn Analytical Model of the Write Process in Digital Magnetic Recording,xe2x80x9d A.I.P. Conference Proceedings on Magnetic Materials 5, p. 738 (1971).
For increased TPI, the media should also have high Hc to support a high off track compressibility (OTC). OTC is the measure of how much the individual tracks can be squeezed together before they start to interfere with each other and degrade the error rate as the track being squeezed. For example, in order to support 10 KTPI, the Hc of the media should be equal to or higher than 2500 Oe. In summary, in order for the media to be able to support high areal recording density, Hc of the media must be as high as possible.
Other parameters of importance for recording performance are overwrite (OW), signal to noise ratio (SNR), and total non-linear distortion (TNLD). Overwrite is a measure of the ability of the medium to accommodate overwriting of existing data. In another words, OW is a measure of what remains of a first signal after a second signal (for example of a different frequency) has been written over the original data. OW will be low or poor when a significant amount of the first signal still remains after over-writing. OW is generally affected by the coercivity, the squareness, and the SFD of the medium. For future high density recording, higher Hc media will be preferred for narrower PW50 and high resolution. However, an increase in Hc is generally accompanied by reduction in OW. Thus, there is a need in the art to improve the S* and the SFD to obtain improvements in OW for a given Hc. High hysteresis loop squareness and narrow switching field distribution can be obtained by having uniform distribution of the grain size of the media.
Another factor which is important for increased KFCI and KTPI is that the signal to noise ratio must be maximized. There are contributions to SNR from the electronics and the channel used to process the magnetic signal. But there is also intrinsic noise from the media that must be minimized. The largest contribution to the media noise is generated from the interparticle (or intercrystalline) magnetic exchange interaction. To suppress this exchange interaction, one must isolate the magnetic crystals from each other by one or more nonmagnetic elements (such as Cr atom) or compounds. The amount of separation need be only a few angstroms for there to be a significant reduction in intergranular exchange coupling. Another source of intrinsic media noise is the size or dimension of the magnetic grain. As the recording density approaches 2 Gb/in2 and beyond, the bit size along the track will be in the order of 0.1 xcexcm or less. Therefore to prevent the excessive noise arising from the physical dimensions of the grain, the diameter of each magnetic grain on the average should be approximately 0.01 xcexcm (10 nm) or less for media in the 2.0-2.5 Gb/in2 media range. The grain size should be approximately 7.5 nm or less for 5 Gb/in2 media, and should be approximately 5 nm or less for 10 Gb/in2 media. Intrinsic media noise has been theoretically modeled by J. Zhu et al. in xe2x80x9cMicromagnetic Studies of Thin Metallic Filmsxe2x80x9d in Journal of Applied Physics, Vol. 63, No. 8, (1988) p. 3248-53 which is incorporated by reference herein. T. Chen et al. also describe the source of intrinsic media noise in xe2x80x9cPhysical Origin of Limits in the Performance of Thin-Film Longitudinal Recording Mediaxe2x80x9d in IEEE Transactions on Magnetic, Vol. 24, No. 6, (1988) p. 2700-05 which is also incorporated by reference herein.
There is another intergranular interaction, called magnetostatic interaction, which acts over a much greater distance between particles as compared to the exchange interaction. Reducing the magnetostatic interaction does reduce intrinsic media noise slightly. However, the effects of magnetostatic interaction actually improve hysteresis loop squareness and narrow the switching field distribution (but to a lesser extent than the exchange interaction), and hence improve PW50 and OW. Therefore, magnetostatic interaction is generally desirable and hence tolerated.
Total non-linear distortion (TNLD) is another parameter that needs to be reduced for high density recording, and it comes about from intersymbol interference between adjacent bits, and partial erasure of a bit at the transition during writing. TNLD can be reduced by increasing the coercivity, reducing the remanent magnetization Mr, and reducing the film thickness T, for a reduction in remanence thickness product MrT. TNLD is also improved by orienting the easy magnetization direction of the magnetic particles in the plane of the media. Increasing the Hc is desirable not only for TNLD but also for PW50 as mentioned previously. Since TNLD increases as the recording density increases, it is becoming an important parameter for higher recording density.
In order to obtain the best performance from the magnetic media, each of the above criteria such as PW50, resolution, OW, SNR, and TNLD must be optimized. This is a formidable task, as each of these performance criteria are interrelated. For example, obtaining a narrower PW50 and reducing TNLD by increasing the Hc will adversely affect OW. A thinner medium also provides narrower PW50, better OW, and lower TNLD, however the SNR decreases because the media signal itself is reduced. Increasing squareness of the hysteresis loop contributes to narrower PW50, better OW, and lower TNLD, but may increase noise due to intergranular exchange coupling and magnetostatic interaction. The SNR of the media can be reduced by decreasing the grain of the media, however smaller grain size may reduce Hc due to onset of super-paramagnetic effect which comes about due to inability of the grain to support the magnetization when it competes with the thermal fluctuation. In general, the effect of onset of super-paramagnetic can be delayed by increasing the Ku of the magnetic grain through addition of platinum which has high orbital moment, and also improving the crystalline perfection of the hexagonal close packed (HCP) cobalt grains.
Therefore, an optimal thin film magnetic recording medium for high density recording applications that can support high bit density will require low noise and high signal without adversely sacrificing PW50, OW and TNLD. The cobalt alloy which is currently used for optimization of certain of the above performance criteria typically includes the addition of chromium (Cr), tantalum (Ta) and platinum (Pt), due to their ability to provide high Hc and high magnetic moment. Chromium is typically added in the amount greater than 10 atomic % to act as segregant to separate the cobalt alloy grains for noise reduction and for corrosion resistance. Other additives such as Ti, V, W, Mo are sometimes used. In all cases, the cobalt crystal structure must be hexagonal close-packed (HCP), and it is preferable to have the c-axis of the grains oriented in the film plane. This is accomplished by depositing a chromium film below the cobalt layer and arranging for the epitaxial growth of cobalt alloy grain above the chromium layer. In order to describe the crystallography of the cobalt alloy and chromium, planes and directions in the crystal are denoted by generally accepted conventions, such as described in xe2x80x9cElements of X-ray Diffractionxe2x80x9d by B. D. Cullity published by Addison-Wesley Publishing Co. Inc., herein incorporated for reference. It is typical to describe the crystallographic planes and directions in hexagonal crystals such as cobalt by a 4 indices notation called Miller-Bravais indices, while cubic structure crystals such as chromium are denoted by 3 indices notation called Miller indices. Brackets, xe2x80x9c less than   greater than xe2x80x9d are used to describe crystallographic directions, while parenthesis xe2x80x9c( )xe2x80x9d are used to denote specific planes. xe2x80x9c{ }xe2x80x9d are used to denote a class of planes which are crystallographically equivalent. For example with chromium with body-centered (BCC) crystallographic structure,  less than 001 greater than  direction is normal to a (001) plane. For a hexagonal crystal structure such as cobalt, the crystal surface with the most dense atomic packing is the (0001) plane and the direction normal to that plane is  less than 0001 greater than  direction. The  less than 0001 greater than  direction is often referred to as the c-axis as described earlier. The crystallographic directions and the surfaces for cobalt is shown in FIG. 1 and those for chromium are shown in FIG. 2.
The crystallographic orientation relationship that occurs between hexagonal cobalt film and BCC chromium film was originally reported by J. Daval and D. Randet in xe2x80x9cElectron microscopy on High-coercive-force Coxe2x80x94Cr Composite Filmsxe2x80x9d in IEEE Transaction on Magnetics, MAG-6, No. 4, (1970) p. 768-73. This work was preceded by work by J. P. Lazzari, I. Melnick and D. Randet in xe2x80x9cExperimental studies using in-contact recording on chromium-cobalt filmsxe2x80x9d in IEEE Transactions on Magnetics, Vol. MAG-5, No. 4, (1969) p. 955-59 where they reported that Hc of the cobalt film is increased by its deposition on top of a chromium underlayer.
The crystallographic orientation of chromium which promotes the cobalt c-axis to lie in the plane of the film is to arrange for chromium film to grow with  less than 001 greater than  preferred growth, which means that {001} type planes of chromium lie in the plane of the film. It has been found that cobalt {11{overscore (2)}0} type planes match well with the atoms on a {001}Cr plane as shown in FIG. 3 hereof. Lattice spacing for pure Cr (ao=2.885 xc3x85) and pure Co (c=4.069 xc3x85, ao=2.507 xc3x85) is illustrated in FIG. 3. As seen in the Figure, the  less than 0001 greater than  direction of cobalt is aligned with the  less than 110 greater than  direction of the chromium lattice in the plane of epitaxy. In this direction, the Cr and Co lattices are closely matched and the mismatch is around 0.3%. Along the orthogonal direction ( less than 01{overscore (1)}0 greater than Co), the mismatch with the Cr lattice is much larger at around 6.4%. In this orientation relationship between cobalt and chromium, the lattice match is close only in one direction. The same holds true for alloys of cobalt. It should also be pointed out that in the above orientation relationship between chromium and cobalt, there are two equally plausible configuration for the cobalt. The  less than 0001 greater than Co direction can lie along two orthogonal  less than 110 greater than Cr type directions. In fact when the chromium grains are large, two variants of cobalt grains which are oriented 90xc2x0 to each other can form on the {001} surface of the chromium grains as described in xe2x80x9cEffect of Microstructural Features on Media Noise in Longitudinal Recording Mediaxe2x80x9d by T. Nolan et al. published in Journal of Applied Physics, 73(10), May 15, (1994) p. 5566-68.
A large variety of cobalt alloys have been used with a Cr undercoat. In its current industrial form, the Cr undercoat thickness is typically between 500 to 2000 xc3x85, and it is deposited on a heated substrate. A high degree of epitaxy between the Cr and the magnetic layer is required in order to obtain high Hc and high hysteresis loop squareness. Typically, Cr grows with strong  less than 100 greater than  orientation above 200xc2x0 C. In the plane of the film, the epitaxial relationship is  less than 110 greater than Cr// less than 0001 greater than Co, and {100}Cr//{11{overscore (2)}0}Co where xe2x80x9c//xe2x80x9d denotes xe2x80x9cparallel toxe2x80x9d. Alloying elements can be added to either chromium and cobalt or to both to attempt to match the lattice better. As mentioned before however, only one direction along the crystallographic direction can be truly matched, while the other direction (orthogonal) will always be mismatched in the above epitaxial orientation.
More recently, it has been shown by K. Hono, B. Wong, and D. E. Laughlin in the article xe2x80x9cCrystallography of Co/Cr Bilayer Magnetic Thin Filmsxe2x80x9d in Journal of Applied Physics 68(9) (1990) p.4734-40 that in-plane c-axis orientation may be achieved through other crystallographic relationship between Cr and Co lattice. The following lattice plane relationships have been proposed: (002)Cr//(11{overscore (2)}0)Co, (110)Cr//(10{overscore (1)}0)Co, (110)Cr//(10{overscore (1)}0)Co, and (211)Cr//(10{overscore (1)}0)Co. Generally, the addition of alloying elements into cobalt expands the lattice. For an alloy composition of CoCr10Pt18 for example, the lattice parameters are calculated to be co=4.148 xc3x85, ao=2.556 xc3x85. For this composition, the lattice mismatch for pure Cr and CoCr10Pt18 alloy for several combination of planes are listed in Table 1. The two mismatch numbers are two orthogonal directions in the plane of the epitaxy. It can be seen from table 1 that best epitaxial match can be obtained between (110)Cr and (10{overscore (1)}1)Co for a lattice mismatch of 0.2% and 2% respectively. However, in this case the c-axis of the cobalt is tilted 28xc2x0 out of the plane. Another closely matched relationship is (211)Cr and (10{overscore (1)}0)Co for a lattice mismatch of 1.7% and 2% respectively. For other epitaxial relationships where the c-axis lies in the plane of the film, the difference in mismatch along the two directions is always greater.
Since the original work by J. Daval and D. Randet on Co/Cr epitaxial film structure, there are many examples of work on both the cobalt and chromium underlayer alloys to improve the recording performance of the cobalt/chromium alloy structure. A variety of schemes have been proposed to improve the lattice matching between Cr or Cr alloys with the cobalt alloy, and hence improve the in-plane orientation of the cobalt, and improve Hc and other properties as previously noted. There are several approaches. The first involves alloy or deposition variations on a basic two layer structure, involving Cr and Co alloy films. The second approach involves use of multiple layers in the undercoat or a different material other than Cr or alloys of Cr in an attempt to affect the magnetic properties. Thirdly, multiple magnetic layers can be used to attempt a better in-plane orientation.
For a basic two layer approach, in an article by Deng et al. entitled xe2x80x9cMagnetic Properties and Crystal Texture of Co Alloy Thin Films Prepared on Double Bias Cr,xe2x80x9d Journal of Applied Physics 73(10) May 15 (1993) p. 6677-79, the authors claim that (200)Cr orientation can be formed on single crystal Si substrates which then allow (11{overscore (2)}0)Co orientation to form on the (200)Cr at room temperature. This is useful in terms of obtaining the good in-plane orientation of the Co, but the Si substrate is considerably expensive compared to conventional NiP/Aluminum substrates or alternative substrates such as glass or glass-ceramic, and therefore it is not practical. The authors also note that with substrate bias, Cr forms a strong (110) orientation, which then causes the cobalt alloy film to form with (10{overscore (1)}1)Co texture which is out of the plane and hence the Hc of the film dropped. Therefore the cobalt alloy film has a natural tendency to form with the out of plane (10{overscore (1)}1)Co orientation on (110)Cr. The Cr thickness was 1000 xc3x85 and Hc wasxe2x89xa62000 Oe for a CoCrTa film thickness of 300 xc3x85. In another paper entitled xe2x80x9cPreferred Orientation in Cr-and Co-Based Thin Films and its Effect on the Read/Write Performance of the Mediaxe2x80x9d by H-C Tsai et al. in Journal of Applied Physics 71(7), Apr. 1 (1992) p. 3579-85, the authors discuss the importance of achieving (200)Cr which then promotes the in-plane orientation of (11{overscore (2)}0)Co for higher Hc and improved parametric performance. On glass ceramic substrates (canasite) however, the authors note that (110)Cr is formed which then causes the Co-alloy to grow with the c-axis out of the plane. In fact a vertical orientation in which the c-axis is normal to the film plane was reported in one case. The authors report that formation of preferred orientation in the Cr layer is determined by substrate material and the oxygen content in the sputtered film. For the in-plane media, a Hc of 1350 Oe was obtained for CoCrTa alloy, and 1370-1830 Oe was obtained for CoCrPtTa alloy. Typical Cr thickness in these films were xcx9c1700 xc3x85, and MrT of the media was also high at around 5 memu/cm2.
Using the second approach, in U.S. Pat. No. 4,652,499 by K. Howard, it is disclosed that to improve the lattice matching between the Co-alloy and the Cr underlayer for better epitaxy, the Cr underlayer is alloyed with Vanadium (V). Other alloying additions such as Ti, Mo, Hf and Ta have been also tried. According to each of these methods, the Cr underlayer must be several hundred angstroms thick to establish the proper Cr texture for epitaxial growth of the Co-alloy. However, the consequence of forming such a thick Cr or Cr alloy layer is that the grain size of the Cr or Cr alloy also grows, to a size of several hundreds or possibly thousands of angstroms in diameter. Since the grain size of the Co-alloy deposited on the Cr underlayer will match the grain size of the Cr underlayer, the resulting epitaxially grown Co-alloy will also have a grain size of several hundreds to possibly thousands of angstroms in diameter. In such a case, the high intrinsic media noise at the magnetic transitions due to the large grains renders the media useless for current and future high density recording.
In an approach using multiple underlayer and different materials, Lee et al. (Carnegie Mellon) in xe2x80x9cNiAl Underlayers for CoCrTa Magnetic Thin Films,xe2x80x9d IEEE Transactions on Magnetics, vol. 30, no. 6, pp. 3951-3 (November 1994); xe2x80x9cEffects of Cr Intermediate Layers on CoCrPt Thin Film Media on NiAl Underlayers,xe2x80x9d IEEE Transactions on Magnetics, vol. 31, no. 6, pp. 2728-30 (November 1995); and European Patent Application Publication EP 0 704 839 A1, claiming priority from a U.S. patent application Ser. No. 08/315,096, filed Sep. 29, 1994, teach a Co based magnetic alloy layer formed above an underlayer having a B-2 crystal structure. The B-2 crystal is a body centered cubic structure alloy such as NiAl. In one embodiment, the Co based magnetic alloy layer is grown epitaxially on the B-2 structure layers directly below. The proposed rationale is that (110)NiAl and (112)NiAl preferred growth orientation of the NiAl layer provides the basis for the Co-alloy to assume a (10{overscore (1)}0)Co orientation, resulting in an in-plane c-axis orientation of the cobalt alloy, and hence resulting in high Hc.
In another embodiment, an extremely thin Cr intermediate layer (25 to 50 xc3x85) is deposited between a B-2 NiAl underlayer and the Co based magnetic alloy layer, in order to increase Hc of the CoCrPt alloy. In this case, the Cr layer is grown epitaxially on the B-2 crystal underlayer, and the Co based magnetic alloy is grown epitaxially on the thin Cr layer. While the Cr layer is necessary for increased coercivity and squareness, the Cr layer cannot interrupt the crystallographic relationship between the (112)NiAl and the (10{overscore (1)}0)Co (i.e., the indirect epitaxy between the B2 layer and the magnetic recording material layer). Thus, Lee et al. require that the intermediate layer be extremely thin (between 25 xc3x85 and 50 xc3x85) to prevent the Cr from developing its own preferred orientation. Lee et al. also state that a Cr intermediate layer is required to prevent contamination of the Co by the excess Al on the NiAl film surface, and that the Cr intermediate layer must be sufficiently thin to (a) minimize diffusion of Cr into the Co film layer, and (b) avoid interference with the epitaxy between the Co alloy magnetic recording layer and the NiAl underlayer. In their examples, relatively high coercivity up to 3300 Oe and squareness (S*) of  greater than 90% are reported for the high Pt content of 18 atomic % alloy.
Despite the high coercivity, high squareness, and fine grain structure taught by Lee et al. (see EPO 704 839 A1), the mere addition of the NiAl underlayer provides only a minimal improvement in the media noise performance, and thus the resulting media is inadequate for current and future high density recording applications. With reference to EPO 704 839 A1, this minimal improvement can be seen in FIGS. 14 and 15 showing carrier noise measurement data, and FIGS. 16 and 17 showing integrated media noise measurement data. Importantly, it appears that the media taught by Lee et al. suffers from high intergranular exchange interaction as evidenced by the high hysteresis loop squareness S* of larger than 90% shown in the above-referenced paper in IEEE Transactions on Magnetics, V.31, p. 2728, (1995). There is no teaching by Lee et al. to address the issue of reducing the noise. In addition, a very thick layer of NiAl of over 1000 xc3x85 is required in order to form the necessary (112)NiAl texture. Such thick layer is expensive to manufacture. Thick layers cause the deposition shields around the cathodes to flake early, necessitating frequent shut-downs for clean-up. Flaking parts also cause defects which is detrimental to the performance of the media. NiAl is also a difficult material to make the sputtering target out of, hence it is expensive to make. Also the use of thick layers necessitates the use of high rate sputtering methods such as using DC magnetron sputtering. A high rate can be used to deposit thick films but the machine must be cleaned frequently since the inside of the deposition chamber becomes coated just as fast as the substrates. If slower deposition method such as RF sputtering is used, the time allowed for each sample must be increased hence the throughput of the machine will be decreased. In either case, the cost of depositing a thick layer makes it impractical and costly for commercial use.
There are several other problems with the NiAl underlayer concept. As it was shown previously in Table 1, a (110)NiAl texture is like the (110)Cr in that it is also likely to cause the growth of (10{overscore (1)}1)Co texture, since the lattice match is equally favorable. Since (10{overscore (1)}1)Co texture has c-axis oriented 28xc2x0 out of the plane, it is not a favorable orientation for good parametric performance. Therefore it is quite possible that under some more varied deposition or substrate conditions that can occur in a typical manufacturing environment, the perfect in-plane orientation of the cobalt alloy layer may not occur. We also believe that the high coercivity taught by Lee et al. results from the use of a relatively large amount of platinum (e.g., xcx9c18 atomic per cent to obtain Hc of 3300 Oe), as opposed to being derived exclusively from the c-axis in-plane orientation of the cobalt and perfection of the crystallite lattice. As will be discussed further below, a properly isolated CoPt based film can achieve a coercivity of over 3,000 Oe, even with less than 13 atomic % Pt in the alloy even without the benefit of Cr epitaxy, (that is without c-axis in the plane of the film) if the grains in the film are properly isolated as we have shown in the copending patent application by Chen et al., Ser. No. 08/802646, which application is assigned to the assignee of the present invention, and which application is hereby incorporated by reference herein. Indeed, such a high coercivity may be produced with random orientation of grains. The fact that an Hc even higher than 3300 Oe was not obtained by Lee et al. with 18 atomic % Pt in the alloy together with the fact that they obtained extremely high hysteresis loop squareness and no improvement in noise performance indicate that the film taught by Lee et al. has significant intergranular exchange interaction between the physical grains.
The third approach to improving in-plane orientation of the cobalt alloy film is the dual magnetic layer approach as taught in many publications as cited below. In a laid-open Japanese patent application JP8-147660 by Yan and Okumura, a double layer of CoCrTa/CoCrPt on Cr has improved recording density and lowered noise. In one example, the total magnetic layer thickness is 200 xc3x85 with thin CoCrPt top layer thickness between 20-60 xc3x85. Cr thickness was 700 xc3x85, and Pt concentration in CoCrPt alloy was between 10-15 atomic %. With CoCrTa:CoCrPt ratio of 150 xc3x85:50 xc3x85(i.e. 3:1), a maximum Hc of 2150 Oe was achieved.
In another laid-open Japanese patent application JP5-120663 by Yamaguchi and Onodera (Fuji Electric), a dual magnetic layer consisting of a CoCrTa first layer and a CoCrPt second layer having low noise and high Hc is described. In this case, the thickness ratio of CoCrTa:CoCrPt between 1:9 to 7:3 is specified, with optimum around 3:7 to 5:5 (half) achieving the combination of high Hc with low noise. For a total thickness of the film at 500xc3x85, the CoCrTa layer is from 50 xc3x85(1/9) to 250 xc3x85(5/5). The maximum Hc obtained was 1900 Oe with a top layer alloy composition of Co82Cr14Pt4 with a 1000 xc3x85 Cr underlayer. The MrT of the media was xcx9c3.6 memu/cm2. The inventors also stipulates that in the preferred embodiment, the CoCrTa film must be sputtered within 15 seconds after the Cr layer is deposited, and CoCrPt is to be sputtered within 10 seconds after the deposition of CoCrTa film. This is to maintain the high Hc of the film and prevent contamination of the film once it has been sputtered.
In a Japanese laid-open publication JP5-109043 by Inao, Utsumi and Kondoh, the order of the magnetic alloy film stack is now reversed, where the first layer above the Cr is CoCrPt followed by CoCrTa alloy. The first layer is 100 to 1000 xc3x85 thick with preferred thickness of 100 to 800 xc3x85. For the second layer, the thickness is 100 to 1000 xc3x85 with a preferred layer thickness of 150 to 700 xc3x85. They claim that it is preferable to have CoCrPt as the bottom layer with its high Hc capability and the low noise CoCrTa film on top for the best performance. In their preferred embodiment, they obtain maximum Hc of about 1700 Oe with 10 atomic % Pt for Cr(3000 xc3x85)/CoPt10 (300 xc3x85)/CoCr12Ta4 (400 xc3x85) structure. The Hc they achieve is actually very low for the Pt content that they use, indicating that less than ideal Ku has been achieved in their media. No mention is made of the orientation of the magnetic media.
In an article titled xe2x80x9cRecording Characteristics of CoCrTa, CoCrPt Double Layerxe2x80x9d by Yamaguchi et al. in 15th Annual Japanese Applied Magnetic Conference (1991), the authors also describe a magnetic double layer consisting of 150 xc3x85 thick CoCrTa alloy as the first layer above the Cr, followed by 350 xc3x85 thick CoCrPt layer as having good resolution, SNR and OW. Films are sputtered by DC magnetron method to achieve Hc of 1200-1900 Oe. Alloy compositions are not specified. In an accompanying article by Kodama et al. in the same journal entitled xe2x80x9cMagnetic properties of CoCrTa, CoCrPt Double Layersxe2x80x9d, the authors show that for 500 xc3x85 total film thickness, there is a smooth transition in Hc from the lower value of 1000 Oe for 100% CoCrTa to higher Hc value of about 2100 Oe for 100% CoCrPt as the ratio of thickness of the two alloy films are changed. Yamaguchi et al. show that for CoCrTa=150 xc3x85 and CoCrPt=350 xc3x85, superior SNR, OW and PW50 are obtained over those of a single layer CoCrPt film.
More recently, a paper by P.Glijer et al. (Univ. of Minnesota) entitled xe2x80x9cAdvanced Multilayer Thin Films for Ultra-High Density Magnetic Recording Mediaxe2x80x9d published in IEEE Transaction on Magnetics, Vol. 30, No 6, (1994), describe the properties of CoCrTa/CoCrPt double layer with the thickness ratio of 50 xc3x85:200 xc3x85. It is claimed that CoCr16Ta2.5 alloy has better lattice match with a Cr underlayer than with CoCr13Pt13 hence leading to better in-plane c-axis orientation of the cobalt alloy magnetic layer. The epitaxial relationship was quoted to be {10{overscore (1)}0}Co matched to {110}Cr and {112}Cr. A Very high Hc of 3720 Oe was obtained for MrT of 0.9 memu/cm2, with a good S* ratio of 0.88. Media noise was expected to be higher for this structure however, due to increased intergranular exchange present in the media.
A paper by L. L. Fang et al. entitled xe2x80x9cNew High Coercivity Cobalt Alloy Thin Film Medium Structure for Longitudinal Recordingxe2x80x9d published in Applied Physics Letters, 65 (24) Dec. 12, (1994) also describes a double layer magnetic structure using CoCr13Ta3 and CoCr10Pt18. In this work, the Cr layer is 1000 xc3x85 and a range of thickness rations of CoCrTa and CoCrPt is explored. For a total double layer thickness of 300 or 400 xc3x85, the Hc peaks at around 4000 Oe for 50 to 60 xc3x85 thick intermediate CoCrTa layer. The orientation relationship is reported as (110)Cr matched to (10{overscore (1)}1)Co. Even though this implies out of plane c-axis orientation of the cobalt alloy, the hysteresis loop squareness is very high. Whereas without the CoCrTa intermediate layer, the magnetic layer shows vertical orientation, as indicated by x-ray diffraction and hysteresis loop. The high Hc obtained by Fang et al. is quite likely due to the high Pt content in the magnetic layer, and the fact that high squareness was obtained despite the out of plane c-axis orientation suggests that media has high exchange coupling.
A paper by B. Zhang et al. entitled xe2x80x9cCoCrTa/CoCrPtTa Double-Layer Films for Magnetic Recordingxe2x80x9d published in IEEE Transactions on Magnetics Vol. 32 No. 5 p. 3590 (1996) also describes recording and magnetic properties of various combinations of double layer structures. In this work, the ratio between CoCrTa and CoCrPtTa layers are varied as in Glijer""s work. When the bottom CoCrTa layer is 1/3 of the total thickness and balance of 2/3 made up of CoCrPtTa, the Hc peaks at around 2500 Oe which is higher than the Hc of each alloy by itself. SNR however decreased monotonically with increasing CoCrPtTa thickness. The hysteresis loop of the double layer film shows a more square loop compared to the single layer films. When the film stacking order was reversed with CoCrPtTa as the bottom layer however, the results were completely different. The SNR remained low for the most part, inheriting the poor SNR of the CoCrPtTa layer, while Hc increased monotonically with CoCrPtTa thickness and never peaking as it did in the reverse stacking. It is claimed that the CoCrTa alloy has the right grain structure, namely smaller and more uniform grain size which can be passed on to CoCrPtTa when it is deposited first above the Cr underlayer. From their results, a thickness in the range of xcx9c60 xc3x85 is required in order to set the proper microstructure of CoCrTa in place for the subsequent proper growth of CoCrPtTa layer. They do not state the orientation relationship in their film, but they show x-ray diffraction data showing a strong presence of (110)Cr along with (10{overscore (1)}1)Co and (10{overscore (1)}0)Co orientations. Alloy compositions are not stated, nor are the deposition conditions in this publication.
Lastly, in an published abstract #CC-09 by J. Zou et al. (Carnegie Mellon) for the Joint MMM-Intermag Conference to be held January, 1998, the authors claim that by interspersing CoCrTa between the (112)NiAl plane and CoCrPt magnetic layer, the (10{overscore (1)}0)Co texture improves and in-plane Hc also increases. (10{overscore (1)}0)Co in CoCrTa is claimed to have better lattice match to the (112)NiAl than CoCrPt. Therefore interspersing a very thin layer of CoCrTa is claimed to cause better in-plane orientation of the CoCrPt film, hence contributing to the high Hc and hysteresis loop squareness.
In all of the aforementioned publications, a common theme is that lattice matching between Co alloy and Cr is claimed as the key contributor to the high Hc and lower noise. Particularly with respect to double layers using CoCrTa first layer and CoCrPt second layer, it is claimed that CoCrTa is better lattice matched to Cr, whether the orientation relationship claimed is (200)Cr//(11{overscore (2)}0)Co, (110)Cr//(10{overscore (1)}0)Co or sometimes even (110)Cr//(10{overscore (1)}1)Co. However, as shown earlier in table 1, a perfect lattice match in all directions is not possible. Furthermore, perfect or even good epitaxy in an actual production machines is also difficult to achieve. One reason is that residual gases in the system which are often water vapor and oxygen which can be produced from the water vapor, can affect the interface between the Cr and the Co alloy layer and cause loss or partial loss of epitaxy.
From our experience in this work, most of the deposition processes reported in the references cited in this application can be considered xe2x80x9chigh-ratexe2x80x9d. Typically rates above approximately 5 xc3x85/sec fall in the high rate category. Although some of the prior art discloses lower deposition rates, such processes are typically used only in research settings, where there is little or no emphasis on throughput. In contrast, it is generally considered desirable to use high rate processes to achieve high throughput in a production environment to make the process commercially viable. For example, in many production level machines, rates as high as 50 xc3x85/sec are typically used in order to maintain high throughput.
One alternative to reducing the effect of residual gases on epitaxy is to improve the vacuum of the deposition chamber. M. Takahashi et al. in xe2x80x9cThe Ultra Clean Sputtering Process and High Density Magnetic Recording Mediaxe2x80x9d published in IEEE Transactions on Magnetics Vol. 33, No. 5 p. 2938 (1997) shows that by maintaining the vacuum at an ultra high vacuum level of about 10xe2x88x929 Torr, higher Hc is obtained at a thinner Cr thickness. It is claimed that high Hc is obtained through increase in the magnetocrystalline anisotropy field Hkgrain, and by a decrease in the degree of intergranular magnetic coupling. The high vacuum allows better segregation of Cr at the grain boundary, and formation of smaller magnetic grains through smaller Cr underlayer grains. These effects also in turn improved the S/N ratio of the media. It can also be inferred that in-plane orientation of the media is improved by reducing the oxidation of Cr underlayer. Although the results are impressive, the type of vacuum levels which are advocated are not practical in terms of a low cost, high throughput manufacturing operation.
Therefore, under deposition conditions that are not entirely perfect in terms of vacuum conditions and at high rate of deposition typically used in a commercial application, obtaining good epitaxy between Cr and Co-alloy layer is difficult to achieve. Hence c-axis in-plane orientation of the hexagonal cobalt film is not necessarily achieved, leading to lower Hc than is possible for a given Pt content of CoCrPt based alloy, and also reduced squareness. In addition, the grain size may not be uniform and grain themselves may not have the proper crystalline perfection for high anisotropy constant Ku. These factors lead to films having less than desirable parametric performance. As it was shown earlier, even under best conditions where epitaxy is favored, there is still inherent mismatch between hexagonal cobalt and cubic chrome lattice. Lack of good epitaxy and subsequent poor growth-of the magnetic layer can result in formation of a large amount of imperfections such as dislocations, stacking faults and other irregularities in the crystalline structure which will reduce Hc potential for given alloy composition and Pt content. If the magnetic crystallites contain a high amount of imperfections, the Ku of the media will be decreased drastically. When the CoCrTa layer is inserted between the CoCrPt based alloy and Cr according to the above cited prior art, it apparently helps the perfection of the CoCrPt based alloy to some extent, but the cited literature still has to resort to high Pt containing alloy to raise the Hc. High platinum costs more, reduces the magnetization Ms of the film and hence requires a thicker film to obtain a given MrT, and also causes more lattice imperfections in the cobalt HCP structure. At above 8% platinum content, there is increased chance of producing FCC (face-centered cubic) crystal structure which will reduce the Ku value of the HCP structure by more than an order of magnitude. Hence adding an excessive amount of Pt in an attempt to raise Hc will defeat the purpose of the benefit of Pt. Because of these difficulties, we have through careful research invented a new and unique method of overcoming the aforementioned difficulties as follows:
In this invention, a thin layer of a Co alloy such as a CoCr based alloy, including an alloy having the same composition used for the main magnetic layer, is sputtered at an extremely low deposition rate of preferably  less than 1.0 xc3x85/second to a thickness of preferably less than 20 xc3x85 on a {200} oriented Cr film under moderate substrate temperatures, followed by high rate deposition of CoCrPt based alloy to create a media which has very high Hc, low PW50, high SNR and low TNLD values. By using this technique, the CoCrPt based alloy layer achieves excellent in-plane crystallographic orientation, and high Hc is achieved with a minimal amount of Pt. The combination of low Pt, high rate deposition of the main magnetic layer and relative insensitivity of the process to the vacuum conditions of the system makes the method very commercially viable. There are also several additional key benefit which the previous literature have failed to achieve. They are as follows: First, the method allows very fine grain structure of cobalt to be formed which contributes to good signal to noise ratio. Second, the fine grain structure combined with chromium segregation between the grains improve the signal to noise ratio even higher. Third, a high degree of in-plane c-axis orientation is achieved in the cobalt layer which provides very high squareness media which in turn helps to improve OW and obtain low TNLD. Fourth, the quality of the grains is high so that high anisotropy constant Ku, is obtained in the magnetic layer, resulting in high Hc without the necessity of addition of excessively high level of Pt. Low TNLD also results from the high perfection of the grains. These results are obtained using very conventional sputtering processes, without having to invoke ultra high vacuum conditions.
Also there is apparently little impact of specific % lattice mismatch between the Cr underlayer and the CoCrPt based alloy in order to obtain high Hc and good recording performance as long as there is a (200)Cr orientation in the Cr to allow some matching with the subsequently deposited cobalt alloy film. Similar results can be obtained for different Cr alloy layers, such as CrV films. There is much more flexibility now to choose the alloy composition based on the ability of the alloy to affect such parameters as ability to segregate, or to form small grain sizes as opposed to adjusting the magnetic and underlayer compositions to adjust for lattice matching between the two as proposed conventionally.
The present invention consists of magnetic thin film media with the structure of chromium or chromium alloy underlayer, followed by an ultra-thin nucleation layer consisting of CoCr based alloy deposited at slow deposition rate with sufficient atomic mobility conditions, such as with substrate heating or in combination with DC bias on the substrate, followed by full deposition of CoCrPt based magnetic alloy at high deposition rate which is the primary layer for the recording. First, a Cr or Cr alloy film is sputtered using preferably RF diode or RF magnetron cathodes, with heating up to around 250xc2x0 C. or lower at a base pressure of 1xc3x9710xe2x88x926 Torr or better. A conventional deposition rate of between 5 to 50 xc3x85/sec is first used to deposit Cr layer to obtain preferably a (200)Cr orientation. The Cr thickness is typically 300 xc3x85 or less. The second layer which is a nucleation layer consisting of preferably a CoCr based alloy such as CoCrTa is deposited at a very low deposition rate, and it is designed to form a small-grained template for the subsequent growth of the CoCrPt based alloy magnetic layer. An important discovery is that the nucleation layer is very thin, at preferably less than approximately 20 xc3x85 thick, and that it should preferably be deposited at a very slow rate of less than approximately 1.0 xc3x85/sec. In addition, sufficient mobility is preferably imparted to the slowly depositing atoms. This can be achieved by various means including applying substrate heating, substrate DC bias, and using an RF sputter deposition process. The RF process allows for low sputter rate and also induces high mobility to the sputter species from the high bombardment of substrate with electrons.
Alternatively, instead of RF diode sputtering, DC magnetron sputtering can be used, particularly in the high rate deposition of Cr underlayer and the top magnetic layer. For the nucleation layer, DC magnetron can be used for low rate deposition, but it will be more difficult to obtain low rates compared to the RF deposition method. If DC magnetron sputtering is used, additional means of providing high mobility to the sputtering species are desirable, such as a higher level of substrate heating and/or higher DC substrate bias.
By forming the ultra-thin CoCr based alloy nucleation layer by the above method, the magnetic properties of CoCrPt based alloy film such as CoCrTaPt sputtered at high deposition rate above the nucleation changes drastically over the case where the nucleation layer does not exist. The Hc is dramatically raised and hysteresis loop squareness increases. This media will typically exhibit coercivity squareness and remanent coercivity squareness of 0.8 or better and a switching field distribution approaching 0.2 or less. At the same time, the intrinsic media noise is reduced substantially, and resolution is enhanced. TNLD and OW values are also improved over the performance of a non-epitaxial three-dimensional isotropic film. The media is ideally suited for current and future high storage density applications, such as use with magneto-resistive (MR) heads and future giant MR head.
An important aspect of our invention is to overcome the imperfect lattice matching between Cr or Cr alloy underlayer and the magnetic layer by introducing a very good lattice matching of a nucleation layer to the magnetic layer. At low deposition rates with high enough atomic mobility of the nucleation layer, a near equilibrium growth condition is obtained so that a near perfect HCP crystal structure of the nucleation layer is obtained. The thin nucleation layer then serves as a template for the high rate deposition of magnetic layer. Since the main magnetic layer has the same crystal structure as, and is very closely lattice matched to the nucleation layer, there is a high likelihood of good epitaxy between these two layers, hence leading to in-plane orientation of the c-axis and more perfect crystal lattice structure of the cobalt alloy grains. Achieving an equilibrium or near equilibrium condition during the initial growth of the nucleation layer on the Cr or Cr alloy underlayer appears to be very critical to achieving good epitaxy between the nucleation layer and the Cr underlayer. The Cr or Cr alloy underlayer has cubic BCC structure on which the hexagonal HCP structure of the cobalt alloy film must grow. The two lattices do not match very well, and near equilibrium conditions with high mobility of the depositing species will favor better epitaxy. The surface of the Cr or Cr alloy film is also very sensitive to vacuum. It can be easily contaminated or oxidized so that even if proper crystallographic orientation of the Cr layer has been deposited, the subsequent magnetic layer may not grow epitaxially above it. In fact, the cobalt alloy film generally does not favor growth with the c-axis in the plane of the film. The (0002)Co growth (perpendicular orientation) is much more favored, especially at a high deposition rate, so that any problem in epitaxy between a depositing Co alloy film and a Cr underlayer will cause perpendicular orientation to form. Near-equilibrium, slow-rate deposition overcomes this tendency for perpendicular orientation in the cobalt alloy film, and also allows time for the depositing atoms to xe2x80x9cfindxe2x80x9d the correct epitaxy for in-plane c-axis orientation. Near equilibrium growth state can also expel the impurity atoms such as oxygen and other residual gas from the surface.
Once a CoCr based alloy nucleation layer is formed, the surface is more favorable for subsequent deposition of the magnetic layer, even at moderate vacuum conditions so that subsequent to its deposition, the magnetic layer film can be deposited at a higher rate and achieve good in-plane orientation as long as it has close lattice matching with the nucleation layer. It was discovered that this can occur even when the nucleation layer film thickness is only few angstroms thick. Therefore the process is well-suited for use in production equipment where the vacuum levels are usually poorer compared to machines typically used in research. In using a typical production deposition tool for making the recording media, a high rate deposition is necessary in order to make the process economical. But in so doing, thermodynamic equilibrium is difficult to achieve and good epitaxy is often compromised. By splitting the magnetic layer process into two parts, namely a thin and slow deposition rate layer and thicker high rate deposition layer, high throughput is still maintained. Therefore the structure and the method of deposition of the layers are well-suited for production deposition process.
In previous art using much thicker films of CoCrTa as an intermediate layer between Cr and CoCrPt, it is claimed that better lattice match between Cr and CoCrTa alloy as opposed to Cr with CoCrPt is the key contributor to obtaining high Hc and in-plane c-axis orientation. In this invention, there is no limitation to CoCrTa as the alloy of choice for nucleation layer. Since it is the slow deposition rate with high mobility which determines the quality of the epitaxy between the Cr underlayer and the nucleation layer, there is more freedom in the alloy selection for the nucleation layer. In fact, the most ideal choice would be that the same alloy composition is used for both the nucleation layer as well as for the magnetic layer. In practice, a small modification from the magnetic alloy can be used for the nucleation layer composition. For example, for a CoCr15Ta4Pt3 magnetic layer, CoCr15Ta4 without the Pt can be used as the nucleation layer to save the cost of the Pt, but the lattice constant of the two alloys are very close to each other (e.g., within 0.5% lattice mismatch). For higher Pt content magnetic alloy, the nucleation layer may require additional Pt so as to maintain a lattice constant that is closer to that of the magnetic layer. It is quite important that there is a close lattice match between the nucleation layer and the magnetic layer in this invention in order to obtain good epitaxy and also to obtain a high quality crystalline film in the magnetic layer for high Ku values. However, since the crystal structure will be the same hexagonal structure between the nucleation layer and magnetic layer, there is actually more flexibility in the choice of alloys for the two layers. Therefore the alloy selection can be based more on factors such as ability to form small and uniform grains or on basis of creating segregated grains as opposed to selection based on lattice matching considerations alone as is the case with previous art. See the earlier referenced patent application of Chen et al., Ser. No. 08/802,646, for a description of some methods of achieving small and uniform grains and for creating segregated grains.
Control of grain growth and structure is also a key benefit of the present invention. Media according to the present invention comprise a magnetic recording layer having fine individual magnetic grains having a diameter of 100 xc3x85 or less which are isolated from one another by a solid segregant such as excess Cr in CoCrPt base alloy. The magnetic grains have a uniform size, and spacing on the order of about 10 xc3x85. Importantly, the grains of the magnetic recording layer are principally single crystals with minimum defects in the grains, such as dislocations and stacking faults throughout the thickness of the layer, so as to maximize the magnetocrystalline anisotropy constant, Ku, of each individual grain, and furthermore, the c-axis of the grains are oriented predominantly in the plane of the film and are randomly oriented in the plane to provide high Hc. Such magnetic grain structure is a consequence of the use of thin nucleation layer formed in slow rate deposition under near equilibrium condition and high atomic mobility. Since the main magnetic layer grains grows epitaxially over the nucleation layer grains, controlling the size of each nucleation site (grains) and the spacing between adjacent nucleation sites will provide an easy means for controlling the grain growth characteristics, size and spacing of the magnetic recording layer, as well as providing large angle grain boundaries. In addition, proper control of the size and spacing of the nucleation layer grains provides a method for optimizing the segregation of segregant material such as Cr at the grain boundaries in the magnetic recording layer.
The consequence of having the microstructure as described above on the magnetic properties of the media are as follows. Smaller grains contribute to lower intrinsic noise in the recording media. More crystalline perfection in the grains provide the highest possible Ku so that high Hc is obtained for a given Pt concentration in the alloy. Uniform and small grains also contribute to improving the OW. TNLD is improved by having the c-axis in the plane of the film. High hysteresis loop squareness is obtained with having uniformly isolated grains, with uniform size distribution.
Lastly, the process used herein is highly economical and reproducible. With even a relatively low platinum content of around 3 atomic %, an Hc value of about 3000 Oe has been obtained at the MrT value of 0.7-0.8 memu/cm2. This is in contrast with the very high Pt content used by others to obtain high Hc in the literature we have cited. Reducing the Pt content in the sputtering target will reduce the cost of producing the disk significantly, and increase the Ms of the film. This means that film thickness itself can be reduced for a given MrT product, which results in better performance through effectively lower head-media spacing.